SSB

[Martin Czech]

Another book of interest dropped into my hand:

Amateur Radio ...

Everyone knows what modulation is, with a "ringmodulator"
you'll get upper and lower sidebands, with the typicall "beating"
sound, if the carrier frequency is quite low.

Now, those radio people always want to get rid of one of the sidebands
for serveral reasons. This is called single side band modulation or ssb.

This could be very intersting for us e.music people and diyers, too.

How to get rid of the unwanted sideband:

-Filter it out with very steep electromechanic 450 kHz, 3kHz bandwith
 filter. These filters have orders about 100, but are not easy to get.
-Use Quartz filters. Less selective, also not easy to get.

-Polyphase method after Weaver et al.
 You need at least two modulators, feed them with signal a (modulator)
 and  phase shiftet a, and signal b (carrier , ~ 10kHz) and 
 phase shiftet b:
                                  upper              lower   sideband

 cos(a)       *       cos(b) = 0.5*cos(b+a)        + 0.5*cos(b-a)
 cos(a+90deg) * cos(b+90deg) = 0.5*cos(b+a+180deg) + 0.5*cos(b-a) 

Adding both modulators output will give lower, substraction will give upper
sideband.

So far theory. But how to get 90deg phase shift ?
For the carrier this is simple: divider by 4 and 1 out of 4 demux
will give pretty 90deg shiftet square waves. Modulation with 
square waves is ok, because if f=10kHz, the next harmonic sidebands
will occour at 30kHz, and we can make those harmonics weaker with filters.
10 kHz is a good choice : not to high, to give some distance between sidebands,
and not to low, the lower sidebands are "folded" and can be reconstructed 
without interference.

But how to shift the modulator, which is our audio input ?
The radio people use a "polyphase" network of about 4*7 Rs and 4*7 Cs,
low tollerance (1%). This should give 90deg phase shift for a 
frequency range of 1:100 with some error of 2deg.

This should in turn give 40dB sideband rejection from 20Hz to 2kHz
(in audio baseband). The rest could be eliminated using a steep Cauer
lowpass or highpass.

Someone on the list did mention a ssb modulator once...
Will 40dB rejection (if it can be ever obtained)  be good enough ?
Will the sound be worth all this effort.
I'm currenty evaluating this idea.

m.c.